Lecture 2: Project Design & Data Visualization

Review of Lecture 1

Covered

• Inductive vs deductive reasoning
• Formulating research questions
• Accuracy vs precision
• Data types and classifications
• Setting up R projects
• Installing and loading libraries
• Reading files into R
• Creating basic graphs

Lecture 2: Project Design & Data Visualization

The objectives:

1. Design a well-organized project
2. Implement good naming conventions
   • Controlled vocabulary
   • Including units in names
3. Create and use metadata effectively
4. Build tidy, well-structured spreadsheets
5. Understand data repositories
6. Create effective visualizations with ggplot2

<>

Project Design: Step 1

• Data: the raw material of science
• Wide variety of formats, sizes, complexity
• Data management and curation often under emphasized
• Good data management: owe it to our funding agencies, colleagues, supervisors, and study systems

<>

Lecture 2: Project Design: Step 1

1. Determine data types you’ll collect
2. Establish controlled vocabulary
   • Example: do_mgl for dissolved oxygen in mg/L
   • Example: drp_ugl for dissolved reactive phosphorus in μg/L
3. Plan your data flow from collection to analysis
4. Organize your project structure (folders, files)
5. Enter data promptly after collection
6. Save in multiple formats (Excel and CSV)
7. Ensure tidy data principles from the start

Note

See Hadley Wickham’s Tidy Data principles for best practices

<>

Project Design: Step 2

Create a Metadata Sheet that includes:

• Variable descriptions
• Units of measurement
• Collection methods
• Instrument details
• Dates and locations
• Any other relevant contextual information

<>

Practice Exercise 1: Pine Data Organization

Let’s examine our pine needle data: - What naming conventions did you choose? - How did you organize the data? - How can you verify data formats (numeric vs categorical)? - What’s your plan for organizing outputs and figures?

# Code to read and examine data
library(tidyverse)
library(patchwork)
library(kableExtra)

pine_df <- read_csv("data/pine_needles.csv")
glimpse(pine_df)

Rows: 48
Columns: 6
$ date    <chr> "3/20/25", "3/20/25", "3/20/25", "3/20/25", "3/20/25", "3/20/2…
$ group   <chr> "cephalopods", "cephalopods", "cephalopods", "cephalopods", "c…
$ n_s     <chr> "n", "n", "n", "n", "n", "n", "s", "s", "s", "s", "s", "s", "n…
$ wind    <chr> "lee", "lee", "lee", "lee", "lee", "lee", "wind", "wind", "win…
$ tree_no <dbl> 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 2, 2, 2,…
$ len_mm  <dbl> 20, 21, 23, 25, 21, 16, 15, 16, 14, 17, 13, 15, 19, 18, 20, 23…

Lecture 2: Data Management: Step 3

Storage and Backup Strategy:

1. Store raw data and metadata securely
   • Save in both Excel and CSV formats
   • Consider write-protecting raw data files
2. Implement the 3-2-1 backup rule:
   • 3 total copies of data
   • 2 different storage media
   • 1 offsite location (cloud storage)
3. Establish a regular backup schedule

<

Lecture 2: Data Management: Step 4

Initial Data Inspection:

1. Examine data in the Environment tab
2. Run summary() and glimpse() functions
3. Create exploratory visualizations
4. Check for outliers, errors, and missing data

# Specify how to handle missing values during import
pine_df <- read_csv("data/pine_needles.csv", 
                    na = c("", "NA", "N/A", "missing", "null"))

# Get a quick summary
summary(pine_df)

Practice Exercise 2: Try plotting a histogram of your data

Create a histogram of pine needle lengths to check the distribution:

# Write your code here to make a plot
# How do you examine the data - what are the ways you think and lets try it!

Lecture 2: Data Management: Step 5

Data Cleaning:

1. Correct errors and inconsistencies
2. Replace missing values with proper NA codes
3. Document all changes made to raw data
4. Save a clean, master version (consider making read-only)
5. Keep notes on data cleaning procedures

Lecture 2: Data Management: Step 6

Analysis and Visualization Workflow:

1. Create exploratory visualizations
2. Summarize and transform data as needed
3. Document all analysis steps
4. Save outputs systematically

A good way to organize script files is number them in the order they get run.

Lecture 2: Effective Data Visualization

Why make plots?

Get in a group and discuss

• What is the purpose of a data visualization?
• What elements are essential in an effective plot?
• What characteristics define a “good” plot?
• What common mistakes make plots ineffective?

Napoleon’s Disastrous Invasion of Russia Detailed in an 1869 Data Visualization: It’s Been Called “the Best Statistical Graphic Ever Drawn”

Lecture 2: Tables vs. Visualizations

How readable are tables?

We will get to what these number mean and how to make them in the next lecture.

• Tables
  • are they useful in a presentation?

wind	n	mean_length_mm	sd_length_mm	min_length_mms	max_length_mm
lee	24	20	2.45	16	25
wind	24	15	1.91	12	19

Lecture 2: Displaying data

• how does a table compare to a plot?

• Does this help?

• What is this plot?

  • if you don’t explain does the audience know?

<

Lecture 2: Principles of Effective Graphics

According to Tufte (2001), good scientific graphics:

1. Show the data without distortion
2. Maximize data-ink ratio (minimize non-data elements)
3. Make large datasets coherent and understandable
4. Encourage comparison between elements
5. Reveal multiple layers of information
6. Serve a clear purpose in telling your story
7. Integrate with statistical methods appropriately

group	mean_length	sd_length	n	se_length	conf_low	conf_high
cephalopods	18.00000	3.861229	12	1.1146409	15.54669	20.45331
crayfish	18.00000	3.861229	12	1.1146409	15.54669	20.45331
salmon	16.33333	3.938928	12	1.1370705	13.83066	18.83601
snail	18.33333	2.269695	12	0.6552045	16.89124	19.77543

Lecture 2: Creating Effective Graphics

According to Tufte (2001), good scientific graphics:

• To implement these principles:
  • Focus on the data, not decorative elements
  • Ensure proportional representation of numbers
  • Provide clear and informative labels
  • Remove unnecessary elements (“chart junk”)
  • Revise and refine visualizations iteratively

Lecture 2: Displaying data

To make good graphics:

• Above all, focus on data
• Do not distort data
• Graphical representation of numbers → directly proportional to numbers
• Strive for clarity through labeling
• Maximize data-ink ratio
  • Remove non-data ink
  • Reduce redundant data ink
• Revise and redraw

Lecture 2: Displaying data

What do you think?

Does this -

• Focus on data
• Distort data
• Is it directly proportional to numbers
• Is labeling clear
• Maximize data-ink ratio
  • Remove non-data ink
  • Reduce redundant data ink
• Revise and redraw

<

# Let's create two versions of the same plot
# First, a "poor" version with low data-ink ratio
library(ggthemes)
poor_plot <- ggplot(pine_df, aes(x = wind, y = length_mm)) +
  geom_bar(stat = "summary", fun = "mean", fill = "lightblue", 
           color = "black") +
  geom_errorbar(stat = "summary", fun.data = "mean_se", width = 0.5) +
  # theme_excel() +
  labs(title = "Average needle length",
       subtitle = "This plot has a low data-ink ratio",
       x = "Sampling Side", y = "Average lenght (mm)")
poor_plot

Lecture 2: Displaying data

What do you think?

Does this -

• Focus on data
• Distort data
• Is it directly proportional to numbers
• Is labeling clear
• Maximize data-ink ratio
  • Remove non-data ink
  • Reduce redundant data ink
• Revise and redraw

What is one of the most common plots you make all the time?

# "Better" version with higher data-ink ratio
better_plot <- ggplot(pine_df, aes(x = wind, y = length_mm)) +
  stat_summary(fun = "mean", geom = "point", size = 3) +
  stat_summary(fun.data = "mean_se", geom = "errorbar", width = 0.2) +
  theme_minimal() +
  labs(title = "Average needle length",
       subtitle = "This plot has a low data-ink ratio",
       x = "Sampling Side", y = "Average lenght (mm)")


better_plot

Lecture 2: Displaying data -the bad

Common Visualization Problems

1. Data distortion:
   • Non-zero baselines on bar charts
   • 3D effects that skew perspective
   • Inappropriate scales
2. Excessive “chart junk”:
   • Too many gridlines
   • Unnecessary decorative elements
   • Redundant information
3. Poor color choices:
   • Too many colors
   • Non-colorblind-friendly palettes
   • Colors that don’t print well in grayscale
4. Misleading representations:
   • Pie charts with too many categories
   • Dual y-axes with different scales
   • Truncated axes without clear indication

<

NULL

Practice Exercise 3: Lets try some plots with pine data first

Lets try to make some basic plots

# Write your code here to make a dot plot or X y plot
# How do you examine the data - what are the ways you think and lets try it!
# what is missing - hwo do you tell the effect of wind?

Practice Exercise 4: OK we are closer but what about colors or shape or fills

Lets try to make some more basic plots

This is free time - we will free code this….

Below are some examples of code you will need for the future

# Write your code here to make a dot plot or X y plot
# How do you examine the data - what are the ways you think and lets try it!
# what is missing - hwo do you tell the effect of wind?

Lecture 2: Introduction to the Grammar of Graphics - ggPLOT

We will learn the anatomy of a GGplot is layers

• ggplot2 uses a layered grammar of graphics approach:
  1. Data: The dataset you’re visualizing
  2. Aesthetics: Mapping variables to visual properties
  3. Geometries: The visual elements representing data
  4. Facets: Splitting visualization into subplots
  5. Statistics: Statistical transformations of the data
  6. Coordinates: The space in which data is plotted
  7. Themes: Overall visual style of the plotWe have aesthetics

Lecture 2: Building a ggplot Visualization

Key Components:

1. Aesthetics (aes) map variables to visual properties:
   • x and y positions
   • color, fill, shape, size, alpha
   • group, linetype
2. **Geometries (geom_*)** determine how data is displayed:
   • geom_point(): Scatter plots
   • geom_line(): Line graphs
   • geom_boxplot(): Box-and-whisker plots
   • geom_violin(): Violin plots
   • geom_histogram(): Histograms
   • geom_bar(): Bar charts
3. Position adjustments control how elements are arranged:
   • position_dodge(): Side-by-side elements
   • position_jitter(): Add random noise to points
   • position_stack(): Stack elements on top of each other
4. Labels and annotations provide context:
   • labs(): Title, subtitle, caption, axis labels
   • annotate(): Add text, shapes, etc.

Lecture 2: Fine-tuning your visualizations:

1. Colors, fills, and shapes:

   scale_color_manual(
     values = c("wind" = "darkblue", "lee" = "darkred"),
     labels = c("wind" = "Windward", "lee" = "Leeward")
   )

2. Coordinate systems:

   coord_cartesian(ylim = c(10, 30))  # Zoom in without dropping data

3. Themes:

   theme_minimal() +
   theme(
     axis.title = element_text(size = 14),
     legend.position = "bottom"
   )

4. Combining plots with patchwork:

   plot1 + plot2 + plot_layout(ncol = 2)

Practice Exercise 4: Creating a Publication-Quality Plot

Create a fully customized plot that would be suitable for publication:

# Create a publication-quality plot
pine_df %>%
  ggplot(aes(x = wind, y = length_mm, fill = wind)) +
  geom_violin(alpha = 0.4) +
  geom_boxplot(width = 0.2, alpha = 0.7, outlier.shape = NA) +
  geom_jitter(width = 0.1, alpha = 0.5, color = "gray30", size = 2) +
  stat_summary(fun = mean, geom = "point", shape = 23, size = 3, fill = "white") +
  labs(
    title = "Pine Needle Length Varies with Wind Exposure",
    subtitle = "Needles on the leeward side tend to be longer",
    x = "Tree Side", 
    y = "Needle Length (mm)",
    caption = "Data collected Spring 2023"
  ) +
  scale_fill_manual(
    values = c("wind" = "#1b9e77", "lee" = "#d95f02"),
    labels = c("wind" = "Windward", "lee" = "Leeward")
  ) +
  theme_minimal() +
  theme(
    plot.title = element_text(face = "bold", size = 16),
    plot.subtitle = element_text(size = 12, color = "gray30"),
    axis.title = element_text(face = "bold"),
    legend.title = element_blank(),
    legend.position = "bottom"
  )

Key Takeaways

1. Plan your data management from the beginning
   • Consistent naming conventions
   • Good organization
   • Regular backups
2. Make your data tidy from the start
   • One observation per row
   • One variable per column
   • One value per cell
3. Create effective visualizations by:
   • Focusing on data, not decoration
   • Using appropriate plot types
   • Following good design principles
   • Customizing for clear communication
4. Master the grammar of graphics to:
   • Build plots layer by layer
   • Communicate patterns clearly
   • Tell compelling stories with data

Next Steps

• Practice creating different types of plots
• Learn to combine multiple plots effectively
• Explore statistical transformations in ggplot2
• Develop a consistent visualization style